"Trevor Wilson" <trevor@[EMAIL PROTECTED]
> wrote in message
news:67gaftF2jg0mhU1@[EMAIL PROTECTED]
>
> "Alan Rutlidge" <don't_spam_me_rutlidge@[EMAIL PROTECTED]
> wrote in message
> news:4811dac0$0$1603$5a62ac22@[EMAIL PROTECTED]
>>
>> So what you are implying is that a transistor tester, capable of
reading
>> current gain and Vbe is beyond the budget or the average electronics
tech
>> and beyond his / her competency to use? Pull the other one Trevor,
it's
>> got bells on it.
>
> **No. That's not what I'm telling you. The problem is acquiring the
sheer
> number of devices, in order to obtain suitable matches. Not only that,
but
> Peter matches his large signal devices at 1 Amp. Few techs bother
testing
> at more than a few ma.
So what's going to happen if one of my output devices fails and there are
no
matching spares?
I guess the whole lot of the output devices would need to be replaced. So
the customer bears the cost of the other undamaged devices and the labour
cost associated with that? Show me the economics in that rationalisation.
>>> **All transistors in the amplifier stages are matched to within 1% for
>>> hFE and Vbe. The resultant number is written on the top of the output
>>> devices.
>>
>> Oh forgive me Trevor. Just lifted the lid on my ME850 Hi-Cap. Lots of
>> 5% tolerance parts around the output devices.
>
> **You may care to note several things:
> * There are lots of 1% and 2% (not 5%) resistors around the output
> devices.
> * 2% resistors typically measure better than 0.5% off the shelf.
> * The Emitter resistors are matched to within 1% tolerance. As are the
> other WW resistors.
> * The only other 5% resistors are in non-critical sections.
>
>
>> Makes perfect sense now. Equip the device with 1% transistors and
>> surrounding circuitry has 5% components.
>
> **Wrong.
In which case it makes even more sense that a service manual outlining
these
critical components be made available.
>
> What a ****ing laugh. Obviously
>> both you and PS haven't heard that saying - "A chain is only a good as
>> its weakest link." Also the WW resistors in the output stage have a
PTC
>> so the resistance will change with temperature / power dissipation.
>
> **Points:
> * The resistors are matched.
> * The resistors all drift by the same amount.
> * The amount of drift is around 300 parts per million/oC.
> * The resistors are rated, at maximum power dissipation, to drift less
> than 1%.
> * At 120 Watts (continuous) the dissipation of the resistors will be
> approximately 0.12 Watts each.
> * Temperature rise will be approximately 3oC under these conditions.
> * At the rated drift, that corresponds to less than 0.01%
> * At 220 Watts (continuous - 4 Ohm load) the dissipation of the
resistors
> will be approximately 0.44 Watts each.
> * Temperature rise will be approximately 15oC under these conditions.
> * That corresponds to less than 0.05% change in resistance.
>
> Of course, under normal use, the fan will keep temperature change to
less
> than that. Additionally, The figures quoted are for continuous power
> levels, not regular music.
>
>
>>
>> Without a circuit diagram it isn't hard to guess that there are low
value
>> (0.51 ohm) resistors in each leg of the emitter circuits of the output
>> devices. Matching or no matching (within reason), the currents flowing
>> in the emitter circuits will be effectively shared between parallel
>> output devices in this type of circuit configuation. It's no rocket
>> science secret. The technique is commonly used by nearly every
competent
>> class A/B SS output stage.
>
> **Not quite. ME connect Collectors to the load. This is a far less
common
> method. However, you are correct: It is not rocket science. Peter has
> never claimed as much.
>
>> Read the Cherry and Hooper book on amplifier design. Published in the
>> early 70's it's essential reading for anyone contemplating SS amplifier
>> design. My God it even mentions the V-FET and MOSFETS.
>>
>> As for ME output device matching. By that I assume all the MJ15026
>> devices in any one channel would have the same "batch test" number and
>> likewise all the MJ15024 devices would have their own?
>
> **Maybe, maybe not. Batch numbers don't tell you much. Peter matches the
> devices, based on their MEASURED parameters, not what the manufacturer
may
> claim.
I wasn't refering to the transistor manufacturer's batch numbers but the
numbers hand written on the top of the devices (presumably by PS or one of
his former employees).
>> I'm not saying PS doesn't attempt to match his devices within reason.
>> Just I think 1% matching is overkill and it is unnessecary if one
selects
>> an output device with reasonably tight specifications to begin with.
>
> **Your assumption would hold water, IF ME amplifiers were conventional,
> high Global NFB designs. They're not. To perform properly, the outputs,
> drivers and pre-drivers MUST be carefully matched.
>
> [Anecdote] Back in the 1970s, when I was service manager for Marantz, I
> noted that several of their high end models, when serviced, had to be
> fitted with carefully matched devices. The matching was around 20-25% of
> Hfe. MUCH closer than regular transistor manufacturer matching. At one
> time, I ran short of matched devices. I attempted to use a set of
> unmatched devices to get my customer's amp going. I could not allow it
out
> of the workshop. Distortion went through the roof. From a typical figure
> of around 0.05% (20Hz-20kHz) at full power, the figure rose to in excess
> of 0.5% at mid-band frequencies. A clearly audible figure. And, I might
> add, that this was a more or less conventional, high global NFB design.
>
In that case, I'd say the basic design had some serious shortcomings.
Cheers,
Alan
BTW, thanks for answering the questions.
|
